Yokozawa, Washimi, Tanaka

### Abstract

We stomped the ground around MCF and MCE chamber with locking IMC to check the resistance for the blasting. However, we could shake the ground around MCF and MCE chamber at most +/- 60~70 um/s and IMC could keep the lock during our stomping. According to IMC LSC feedback signal to the laser PZT (K1:IMC-SERVO_SLOW_DAQ_OUT_DQ), when the ground shaked +/- 60~70 um/s, the amplitude of the feedback signal became +/- 0.2 V at 1 Hz (which is the length resonance frequency of Type-C sus.). If the blasting shake the ground with the amplitude of 200 um/s, which is estimated by the construction company, the feedback signal is expected to become +/- 0.6~0.8 V. This value is in the range (+/- 5V) so IMC will be able to keep the lock during the blasting. At worst, IMC lock can restore by guardian automatically even if IMC goes down.

### What we did

• Fig. 1 is the view of the location of the seismometer and the stomping
• With locking IMC, firstly, we stomped the corners of the MCF ENGIRI as following order (see the movie). However, we could shake the ground with the amplitude of at most +/- 60 ~70 um/s by our stomping. This is the one third smaller than the estimated amplitude of the ground motion by blasting. And then IMC kept the lock during the stomping.
  1. -X, +Y (nearby MCi camera, we disconnected the LAN cable during our stomping. After our work, we reconnected the LAN cable.)
  2. -X, -Y (nearby MCi oplev)
  3. +X, -Y (nearby MCo oplev)
  4. +X, +Y (nearby MCF seismometer)
• Then, we moved the MCE area and stomped the MCE ENGIRI corners with the same procedure as the MCF trial. Simirally, IMC kept the lock.
• According to IMC LSC feedback signal to the laser PZT (K1:IMC-SERVO_SLOW_DAQ_OUT_DQ), when the ground shaked +/- 60~70 um/s, the amplitude of the feedback signal became +/- 0.2 V at 1 Hz (which is the length resonance frequency of Type-C sus.) (Fig. 2).
• If the blasting shake the ground with the amplitude of 200 um/s, which is estimated by the construction company, the feedback signal is expected to become +/- 0.6~0.8 V. This value is in the range (+/- 5V) so IMC will be able to keep the lock during the blasting. At worst, IMC lock can restore by guardian automatically even if IMC goes down.

I redesigned the MN_NBDAMP_DOF5 filter not to oscilate at LOCK_ACQUISITION state.
At 296K, ETMX forth L resonance is about 5.04 Hz while it was 5.13 at 90K.
So, I made new filter (BP5.04(296K)) at FM1 of DOF5 and moved old filter from FM1 to FM2.
After changing the filter, damping seems working well (fig1).

Figure 2 show the one hour trend of PS DAMP signals at LOCK_ACQUISITION state after changing the filter.
For one hour, no oscillation can be observed, so the control seems stable.
 

Note that the TFx and TFy are the vertical -> horizontal coupling in this shaking test, but were the horizontal -> horizontal coupling in the previous hammering test. So their comparison is not fair.

HWP vs IMC trans power relationship

I took the picture 21st May. (Tuesday)

[miyoki, uchiyama, hayakawa, yoshimura, yamaguchi, omae, takahasi, sawada]

Progress

We filled the cleaning water just above the bottom surface of the optical table. We left it for one night.

Cleaning Process

1. Water preparation
   • We prepared water filtered by cascading 25um and 5um filters from the tap water obtained at the EY area.
   • We also prepared hot water by using the electrical pot from this filtered water.
   • We once put this water in poly-tanks and mixed it with the hot water to make warm water, then moved the poly-tanks to the EYA area.
   • We made the "cleaning water" by diluting the neutral detergent liquid with 50 times amount of warm water.
2. Absorber setting
   • Because I have set two of three thin red tubes to absorb the water between the bellows and the poles for the optical table, I set the last one for the space at +x/+y side.
3. Bubble shield setting, etc (Fig.1, 2, 3)
   • We inserted sponge bars that were wrapped with stretch seals in the space between the optical table and the side wall of the EYA vacuum tank to reject bubbles from the cleaning water when we injected air for mixing the cleaning water.
   • We put vectra alpha cloths on the mirrors for the P-cal. We are very sorry but one of mirrors seemed to get a liquid drip on the left surface of the mirror. We tried to clean this drip by acetone and apparently, the surface seemed to be clean. However, anyway, please check and clean all mirrors after the EYA tank cleaning.
4. Cleaning wroks
   • This cleaning water was transferred from the poly-tank to the bottom of the EYA vacuum tank by a pump.
   • During the transfer of the cleaning water, we also injected air by using an air pump to mix the cleaning water in the EYA tank. (Fig.4)
     • I tested the water absorber for the bellow-pole space, and I confirmed that water could be absorbed.
   • We stopped the cleaning water injection when the water surface was over the bottom surface of the optical table.
   • I washed the bottom surface of the optical table with a silicon brush that had a long handle. However, the most outside area of the optical table could not to be reached. I will try washing with a wiper that has a longer handle again tomorrow. (I have already bought a longer brush in DAISO)
   • Because the time was up, we left the cleaning water inside the EYA tank.
   • To avoid too much moisture inside the EYA vacuum tank, we left two side hatches to be open and one side mouse on the Y-arm axis on the EYC side was closed. So the air from the FFUs is expected to flow from the -x side to +x side.
   • (When the water surface was approaching the bottom surface of the optical table, the bubbles came out through several holes in the optical bench because these holes completely penetrated the optical bench. So we stopped the bubbling. As we expected, the sponge bars could prevent these bubbles from going on to the optical table surface.)

Tomorrow plan

1. Wash the bottom surface of the optical bench again by using a longer brush.
2. Drain the water inside the EYA tank. (The abandoned water should be transferred again into poy-tanks because we cannot abandon this used cleaning water in the mine water routes.)
3. Inject the warm water again and wash the bottom surface of the optical table again.
4. Drain the water inside the EYA tank and check inside by using the 3D camera.
5. ...

I checked the TFs measured in air. They are consistent with the reference (measurement before O4a) and look healthy.

Abstract

Details

Over the last few days I have been trying to understand the reason for the instability of the 30 mHz blending strategy along the T direction. As already mentioned, the T TF still shows the phase lag visible in the TFs measured with the blended sensor, despite the phase compensator implemented to compensate for it. As with the EX, I first modified the phase compensator to avoid DC saturation of the ACC and GEO (see Figure 1, Figure 2, Figure 3, and Figure 4), and then measured the TFs: LVDT/IS{L,T}.

Figure 5 and Figure 6 show the TF: LVDT/IS{L,T}. It is clear that there is still a phase lag which introduces instability into the loop. To reduce the phase lag, I implemented a new phase compensator on the virtual inertial sensor. I then re-measured the TFS and it seems that it helps to reduce the phase lag and should stabilise the loop (see Figure 7, Figure 8).
Next step:
Test the stability of the loop and redesign the blend filters.

I calculated the transfer functions from the base vibration (z) to the table vibration (x,y,z).

Comparing the results of hammering (klog29515), inconsistency is found.

The underestimation below 70Hz is solved.

I performed the shaker injection tests for the OMC base plate, by locating a 3-axial accelerometer (TEAC710Z) on the optical table and a 1-axial accelerometer (TEAC710, for vertical) on the base plate.

I performed actuator and center balancing of ETMY to confirm strange MNV TF can be better by sensor/actuator decoupling.
Figure 1 shows the MNV TF after decupling.
MNV TF becomes healthy.

Also, I measured TFs from V1 and V3 coils (fig6: V1, fig7: V3).
Since the gain of V1 becomes smaller, the gain of TF is smaller than before but it is not problematic.
Also, gain of MN V3 is now same as the reference, so the smaller gain, which was measured previously, seems due to the gain change in DGS.

So, MN stage TF seems fine now.

What I did:
1. Photosensor gain (MN_OSEMINF_{H1,H3}_GAIN) was changed to minimize the coupling between MNV and MNP (fig2:before, fig3:current).
2. Actuator balance was performed for reducing V2P coupling (fig4: before, fig5:current).
3. V2Y actuator decoupling was performed (fig6)

Is this actual number of the pressure inside IFI-IMM-PRM chambers?? I wonder if the GV between this CC-10 and the IFI-IMM-PRM chambers might open ot not.

[Kimura]
The serial communication settings on the replaced CC-10 were reset to factory settings, but the connection to the network was not restored.
Therefore, the electronic board of the CC-10 was replaced with the removed CC-10 electronic board and the sensor calibration curve was reset.
 As a result, the connection to the network was restored. (Figure 1)
The values were confirmed to be consistent with the displayed values seen by the network camera. (Photo 1)
The CC-10 with communication failure will be sent for repair.

At 11:16 a.m., CC-10 indicated 8.0 x 10^-5 Pa.

Around 9:00, 8.2x10^-5 Pa.

Around 10:00, 1.0x10^-4 Pa.

Abstract

Details

I forgot to mention it.
During this work we found a air-wirering of RF components around IOO0 rack as attached. Such a connection should be avoided.

Aoumi, Kamiizumi, Tomura (mine), Takano (remote)

Abstract

We investigated the oscillation of the common mode servos installed in ALS1 rack one by one. The servos oscillated around 28 MHz as we expected. We also found an extremely large 50MHz peak from Summing node, which seems to come from GrPDH X/Y servo.
We also confirmed that GND of the input signal of FIB X/Y servo is well isolated from GND of FIB X/Y servo (above 1MΩ), that is the reason why these two servo is not oscillated.

Detail

From the previous measurement we confirmed that the common mode servos installed in ALS1 rack (PLL X/Y, CARM, Summing node) oscillate at some MHz. To identify which servo oscillates at which frequency, today we investigated the situation of the oscillation one by one.

Figure1 shows the spectrum with all the servo installed in the rack turned on measured by the same way in the previous measurement. We saw oscillation peaks at 17 MHz and 28.MHz. After turned off all these servo, we saw the spectrum shown in Figure2.

After that, we turned on each servo one by one. Figure3 and Figure4 show the spectra with PLL X turned on and PLL Y turned on, respectively. For PLL X a large peak exists at 28 MHz, which is likely to come from the oscillation of the servo. On the other hand, for PLL Y we ccouldn't see any peak around there. Then, we checked the length of the BNC cable between PFD and PLL X/Y servo, and found that the length is different between X and Y; 1m for PLL X and ~ 2.5m? for PLL Y (it was hard to confrim the actual length, but at least longer than 1m). Therefore, it seems that for PLL Y the length of the BNC cable is enough long not to oscillate.

Next, we chacked the signal from CARM servo and Summing node. FIgure5 shows the spectrum from CARM, and Figure6 is from Summing node. We found a large peak around 28 MHz, which seems to come from the oscillation due to the connection between Qmon of a I/Q demodulator and the servo. For Summing node no peak apeared around 28MHz, but an extremely large peak existed at 50 MHz. It was also confirmed that even if we turned off the power of Summing node this peak still exsists. Therefore, we suspected that this peak comes from Gr PDH servos.

We move on to Gr PHD X/Y servos. We measured the signal around SLOW OUT of each servo, which goes to the inputs of Summing node. The measured spectra are shown in Figure7 for PDHX and Figure8 for PDHY, respectively. It is obvious that there exists a peak at 50 MHz, and this implies that the peak in Summing node comes from Gr PDHX or Gr PDHY. But we are not sure of the source of the peak.

Finally, we checked the GND level of the input of FIB X/Y servos, which don't oscillate from the previous measurement. We measured the resistance between GND of the input BNC cable and GND of OUT2. The results are shown in the table below:

From the results it is clear that GND level of the input of these servos are well isolated from that of the servos, and that is the reason why these two servo don't oscillate.

Conclusion

We almost understand the condition of the oscillation; the common mode servo oscillates if the one of the input is close to GND level of the servo (≈ single ended signal from a circuit with the common GND level) and the cable length is enough short (<1 or 2m?).

On the other hand, we found a large peak at 50MHz existing in Gr PDH X/Y and Summing node. We don't know its source and should investigate it for future.